I am in no way doubting what you are saying - I don’t know what you know. I wanted to share some experience. I have had times where I have seen a large change in results even though no kit at all has actually changed, and realising this means that when comparing different kit I have to be very careful. I reset things several times and take multiple measurements now before changing anything, and then make my change, then multiple measurements afterwards. When I have got multiple results, with numbers that are speeds I take the highest speed, the max not the average, in most circumstances. For other kinds of numbers it’s either the geometric mean or arithmetic mean of the the set of results usually, depending on what I’m dealing with.
The two downstream speeds you quoted need to be adjusted so that they are at the same downstream SNRM, and that would eat up some or all of the speed difference, if I am reading it correctly. Is that downstream SNRM difference 6.2 dB - 5.8 dB = 0.4 dB ? I don’t know how much of that sync rate difference that 0.4 dB would be responsible for. I am also noticing the difference in line attenuation 13 - 12.1 dB - that is very interesting, not sure how to interpret that, but someone who knows what they’re talking about might like to look at the bits-per-bin assignments (tones) to see if there is a possibility of additional insight.